This model brings out the specific styles from the title
V1 was my first ever LoRA and trained on a very small dataset but still effective especially in SFW.
V2 is trained on 1000 images and available on multiple base models. Enhanced NSFW capabilities but retains its SFW strengths.
I recommend using weight/strength of 0.5-0.9 going higher than that sacrifices creativity and makes results look very similar.
You can mix multiple styles (goth, emo, punk, alt, scene) up and use them as keywords but the more descriptive you are the better the results. Like instead of "skirt" you could say "goth skirt" or "punk skirt" as well as "colored emo hair" or "pastel goth hair".
Description
The images in showcase all include my Lain LoRA in metadata but it's a mistake. It has been "used" at 0.00 weight in ComfyUI, so it's not actually doing anything but it still sneaked its way in the metadata.
FAQ
Comments (4)
Reading from the book of souls can bring forward spirits that will give you insight of the dead or a demon that will turn your life into hell!
<3
The SD3.5M lora seems to be treated as SDXL lora in the draw things app - I suppose it might be some issue with the app, but any chance the uploaded safetensor file is the wrong one?
Hi, the SD3.5M Lora is for 3.5 Medium. It was however uploaded before CivitAI remade the classifications to distinguish between Medium and Large LoRAs for SD 3.5 (there used to be only one "type" although there were 2 base models). So maybe that's why something is mixed up. I checked again and the .safetensor is definitely the correct one
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.